OBJECTIVETo examine the main effect of 10 Peroxisome proliferators-activated receptor (PPAR) SNP in contribution to non-HDL-C and study whether there is an interaction in the 10 SNPs.

METHODSParticipants were recruited within the framework of the PMMJS (Prevention of Multiple Metabolic Disorders and Metabolic Syndrome in Jiangsu province) cohort-population-survey, which was initiated from April 1999 to June 2004, and 5-year follow-up data from total 4 582 subjects were obtained between March 2006 and October 2007. A total of 4 083 participants received follow-up examination. After excluding subjects who had experienced stroke or exhibited cardiovascular disease, type 2 diabetes or a BMI <18.5 kg/m(2), a total of 820 unrelated individual subjects were selected from 3 731 subjects on October of 2009. Blood samples which were collected at the baseline were subjected to PPARα, PPARδ and PPARγ 10 SNPs genotype analysis. Logistic regression model was used to examine the association between 10 SNPs in the PPARs and non-HDL-C. Interactions within the 10 SNP were explored by using the Generalized Multifactor Dimensionality Reduction (GMDR).

RESULTSA total of 820 participants (mean age was 50.05±9.41) were included in the study and 270 were males and 550 were females. Single-locus analysis showed that after adjusting gender, age, smoking, alcohol consumption, physical activity, high-fat diet and low-fiber diet factors, rs1800206-V and rs3856806-T were significantly associated with higher non-HDL-C levels. V allele (LV + VV genotype) carriers of rs1800206 have a average non-HDL-C levels on (3.15 ± 0.89)mg/L (F = 15.01, P = 0.002); T allele (CT+TT genotype) carriers of rs3856806 have a average non-HDL-C levels on (3.03±1.01) mg/L (F = 9.87, P = 0.005). GMDR model analysis showed that after adjusting the same factors, two-locus model, five-locus model, six-locus model and seven-order interaction models were all statistically significant (P<0.05), and the seven-locus model (rs1800206, rs3856806, rs135539, rs4253778, rs2016520, rs1805192, rs709158) was the best model (P = 0.001), the cross-validation consistency was 10/10 and testing accuracy was 0.656.

CONCLUSIONRs1800206 and rs3856806 were significantly associated with non-HDL-C. And there was an gene-gene interaction among rs1800206, rs3856806, rs1800206, rs135539, rs4253778, rs2016520, rs1805192, rs3856806 and rs709158 which could influence the non-HDL-C levels.